Abstract
Pyogranulomatous panniculitis due to infection by Mycobacterium smegmatis was diagnosed in two cats in Finland, a country with a rather cold climate. The diagnosis was confirmed by sequencing of the 16S rRNA gene, which gave a perfect match with the M smegmatis strain ATCC 19420. Gene sequencing makes it possible to distinguish M smegmatis from closely related mycobacteria such as M goodii sp.nov. Diagnosing this entity seems to be a question of having a high index of suspicion. The appearance of the disease as well as sampling is described in detail. In our first case an initial erroneous diagnosis of Nocardia species considerably delayed our arriving at the right diagnosis. The first patient has now been followed for more than 7 years. Her disease is chronic, but she is not systemically affected. Several antimicrobials were tried. Probable side effects of enrofloxacin medication are described.
Mycobacterium smegmatis is an opportunistic bacterium, which is considered to be widely distributed in nature (Kamala et al 1994, Tsukamura 1976). It is capable of causing a rare, but well documented pyogranulomatous dermatopanniculitis in cats (Wilkinson et al 1982, Wilkinson & Mason 1991, Malik et al 1994, 2000). This disease has proven difficult to treat. The inguinal fat pad is one predilection site for the infection caused by M smegmatis. Wilkinson, in Australia, first described this condition in cats in 1982. Most reports of this entity come from Australia, where it seems to be a well recognised problem, the largest material being the microbiological characterisation by Malik et al (2000) of no less than 40 strains of M smegmatis isolated from cases of clinical panniculitis in cats. From outside Australia there are only a couple of cases of M smegmatis in cats (Kunkle et al 1983, von Weber et al 2000). Initially it seemed that M smegmatis had clinical relevance only in warm and humid areas, but newer reports seem to contradict this (Malik et al 2000). This report of two Finnish cats further confirms that M smegmatis should also be suspected in colder climates.
M smegmatis is considered to be an organism of low virulence. The first reported human infection was a pleuropneumonia complicating exogenous lipoid pneumonia reported by Vonmoos (1986). Since then at least 26 human cases have been reported, many of which are wound infections in warm climates (Wallace et al 1988, Newton et al 1993, Newton & Weiss 1994). In the records of The National Public Health Institute of Finland there are no reports of M smegmatis causing clinical human disease in Finland in the past 20 years (Kokki, personal communication). And according to the documentation available at the National Veterinary and Food Research Institute of Finland it has, at least during the last 12 years, not been identified in clinical samples of domestic animals in Finland (Seppänen, personal communication).
M smegmatis differs microbiologically only little from some other species of mycobacteria. It has been confused with M avium (Tsukamura 1976) and M fortuitum (Vonmoos et al 1986) and also with a species of Nocardia (Newton et al 1993). Diagnosing has become even more complicated, as M smegmatis has recently been taxonomically subdivided into three distinct groups (Brown et al 1999): M smegmatis sensu stricto, M goodii sp. nov. and M wolinskyi sp. nov. These are microbiologically very similar but have different patterns of antibiotic susceptibility and differ genetically from each other.
Case 1. (a) Two weeks after first presentation: profound subcutaneous swelling with several fistulas. The arrow marks caudal left nipple. X marks biopsy site. (Yellowness is due to artificial light.) (b) Eight months after first presentation: hair had grown back. However, scar tissue could still be felt in the subcutis in the inguinal area and along the left inner thigh. The arrow marks caudal left nipple.
Case reports
Case 1
A two-and-a-half-year-old, 4 kg, spayed domestic shorthair cat was presented in December 1993 with three small puncture wounds in the skin around the left caudal nipple. There was thickening and firmness over a well-defined area of 2 × 3 cm around the nipple. The course of the disease was consistent with panniculitis caused by atypical mycobacteria. The cat failed to respond to oral amoxycillin. During the following week, additional small holes appeared in the skin near the first wounds and there was subcutaneous thickening and firmness that spread in three directions from the central plate. The width of these firm areas was around 1 cm, and the longest extension was several centimetres long, extending over the ventral midline. There were no fistulas at the end of this extension. At this location a piece of dermis with accompanying subdermal tissue was excised in an aseptic manner (Fig 1a). The piece was then divided, and the cut surface was cultured in house on routine blood agar, in 37°C. One piece was sent fresh to a veterinary laboratory with a request for mycobacterium culture. From the other piece imprints were made on slides for Ziehl-Neelsen staining, before it was put into formalin for histopathology. The cat was started on injectable enrofloxacin 25 mg/day (Baytril®, Bayer). On the following day the laboratory confirmed that there were acid-fast bacilli seen on cytology. The histopathology was evaluated later, mainly to check for signs of tumour or foreign bodies. Such could not be found. There was no growth within 24 h on the agar cultured in house, but after a couple of days at 37°C there was a pure growth of identical smooth grey colonies of medium size. This agar plate was then also sent to the same veterinary laboratory for colony identification.
The treatment protocol, for cats with panniculitis by atypical mycobacteria, presented by Malik et al (1994, 2000) had not been published at that time. It involves excising the whole infected area using doxycycline or fluoroquinolone preoperatively for several weeks, gentamycin parenterally intra-and perioperatively for some days and then again mainly doxycycline or fluoroquinolone for 3 to 6 months postoperatively (based on susceptibility data). This regime appears to be the most successful one published so far, with as many as 15 cases of confirmed panniculitis by M smegmatis cured without recurrence even after a long follow-up. Also for human cases aggressive debridement of all infected subcutaneous tissue and skin is generally needed (Friedman et al 2001). Since the protocol by Malik et al was not published at this stage, and the available literature contained warnings for large non-healing wounds after surgical treatment of atypical mycobacterial skin diseases, we hesitated doing surgery and medical treatment was chosen while waiting for the final mycobacterial typing. From day 6 onward the cat was administered enrofloxacin 50 mg/day per os (Baytril®, Bayer). The situation was complicated by the fact that the bacterium was hard to identify. The referral laboratory diagnosed Nocardia spp on the blood agar. Besides this, at that time the laboratory was not able to give any antibiotic susceptibility data for slowly growing bacteria. After 2 months on enrofloxacin alone, 1 tablet of 20 mg trimetoprim and 100 mg sulfadiazine/day (Tribrissen mite®, Schering-Plough Animal Health) and 1 mg folic acid twice/day (Folvite®, Wyeth Lederle) were added to the regimen, because of the erroneous diagnosis of Nocardia. After further 3 1/2 months the compact pad around the nipple had enlarged to 5×8 cm, and there were draining sinuses as far as at the left knee. From the biopsy taken, the mycobacteria department eventually identified a M avium-intracellulare type bacterium, based on gas chromatography conducted at a human hospital. As all feline M avium cases reported in the literature had either before or shortly after diagnosing been fatal, with disseminated spread of infection in internal organs (Buergelt et al 1982, Drolet 1986, Jordan et al 1994), the thought of excising the lesion was abandoned and anti-mycobacterials were now chosen to accompany the TMS. Thus the cat was put on a combination of rifabutin, (Rifabutin®, Pharmacia), clarithromycin 48 mg bid (Klacid®, Abbott) and TMS (the same protocol as before). Blood samples were drawn one to three times a month to check for adverse reactions. Rifabutin was started at a dose level of 7 mg tid, during which the skin healed quickly. The dose was successively reduced, first to 6 mg tid and then, after 6 weeks, to 5.5 mg tid, as serum creatinine increased and the packed cell volume dropped. Liver enzyme activities stayed well within reference range throughout the medication. After 2.5 months the firm subcutaneous pad had reduced in size considerably, but the subcutaneous tissue in the inguinal area still felt abnormal on palpation (Fig 1b). It seemed as if a dose of 5.5 mg tid was not sufficient, as occasional fistulas appeared at the end of this period. Since the packed cell volume had now dropped to 19%, serum creatinine and urea were elevated, and the cat had continuously lost weight during the past 6 months, rifabutin and TMS was withdrawn and clofazimine, 20 mg sid (Lamprene®, Ciba Geigy) was given instead, with the clarithromycin. The cat then gained weight 0.5 kg in 1.5 months, was more active, and the affected blood values returned to normal range, but the panniculitis started to flourish again. During the next 2 years the cat was administered different combinations of clarithromycin, ethambutol, clindamycin, clofazimine and rifabutin (4–5 mg bid). She remained active, but the panniculitis slowly continued to occupy new areas. At worst, about 30% of the body was affected. To intervene more aggressively new biopsies were taken in order to get antibiotic susceptibility tests done on the microbes. Samples were sent, among others, to a human laboratory (United Laboratories Ltd) with a special interest in mycobacteria. This seemed to be conclusive. Both the mycobacterium in the piece of tissue and the bacterium growing slowly on blood agar were identified as M smegmatis. No other acid-fast bacteria were found in two consecutive samples either. This time we received a more complete histopathologic report. Deep in the dermis and especially in the subcutis a diffuse pyogranulomatous panniculitis and dermatitis was seen. On Ziehl-Neelsen stained slides no acid-fast bacteria could be discovered, but with Fite-Faraco stain occasional single bacilli could be discovered centrally within the pyogranulomas. The laboratory confirmed the identification of M smegmatis by sending it for sequencing of the 16S r RNA gene, (the antibiotic susceptibility results are listed in Table 1). When M smegmatis finally was identified the lesions were widespread to a point where surgery could not be considered. After a period of 15 months on enrofloxacin (50 mg/day), rifabutin (4.5 mg bid) and TMS (the same protocol as before) the cat has managed without medication for nearly 2 years without any draining fistulas. However, the subcutaneous fat is firm and bumpy over the lumbar area, the left hindquarter and on the tail, up to 10 cm from its base and the skin has bluish impressions (Fig 2a). Remarkably, the areas affected at the beginning of the disease, such as the inguinal and ventral areas, the left axillar area and the left thoracal wall are well healed without any scar tissue (Fig 2b).
Case 1. Seven years after first presentation. The cat had been off medication for 2 years. While the cat was chronically affected, there had been no active draining fistulas for over 2 years. (a) The subcutis is thickened and bumpy and the skin has dark impressions at the back, in the right lumbar area and on the base of the tail. (b) The inguinal area, the left inner thigh and the left flank, which are the areas originally affected, look and feel normal and smooth.
Antibiotic susceptibilities for case 1 (at 2.9 years, 3.4 years and 4.2 years) and for case 2 (at diagnosis)
The inhibition zones of doxycycline (66 and 70 mm), sulfa (50 and 65 mm) and trimethoprim-sulfa (70 and 70 mm) were large.
Amoxycillin-clavulanate had inhibition zones of only 29 and 37 mm on first and second culture respectively, and on the third culture the bacterium was resistant to this antibiotic, the cat had then been on it for 9 months.
Case 2
Five years later a second case was presented. This was a 6-year-old, overweight, castrated, domestic shorthair cat (7.8 kg), with a 3-month-old infection in the subcutaneous fat caudal to the left axilla. Previous treatment with cephalexin had been unsuccessful. An area of 7×10 cm was affected in a most typical way, with the subcutaneous tissue thickened, firm and attached to both the skin and the underlying tissues. The skin over the affected area had numerous draining fistulae. Purple depression lines of very thin skin ran between adjacent fistulas. The appearance was identical to the advanced stages of our first case. On biopsy the laboratory found M smegmatis with matching 16S rRNA gene sequence to the first case. The antibiotic susceptibilities for case 2 are shown in Table 1. Histopathologically a diffuse pyogranulomatous inflammation was present all the way through the deeper parts of the dermis and the subcutaneous fat to the muscular layer. As in case one no acid-fast bacteria could be seen in slides stained with Ziehl-Neelsen, but single positive organisms were seen with Fite-Faraco. Unfortunately it was almost impossible to administer oral medication to this cat. As the owner had medical training, enrofloxacin (Baytril®, Bayer) was injected at home (Baytril 50 mg/ml 0.6 ml? sterile 1 ml sc) for several weeks, but with time also this turned out to be impossible. Because of the cat's uncooperativeness we anticipated problems with a postoperative medication and the cat was never taken to surgery. Instead the owner elected euthanasia.
Identification by DNA sequencing
From each growing colony a portion was taken into sterile Eppendorf tubes and 100 μl sterile water was added. The tubes were boiled for 10 mins, spun for 5 min and 10 μl was taken for PCR amplification of the 16S rRNA gene as described by Koukila-Kahkölä, 1995. The obtained PCR products were sequenced by using BigDye terminator chemistry and analysed on an ABI 377 automated DNA sequencer. The sequences from both cases were the same and a search through Genbank revealed a perfect match to the sequence for M smegmatis ATCC 19420 (accession number AJ131761).
Discussion
Our observations confirm the former assumptions that pyogranulomatous dermatopanniculitis caused by M smegmatis does not occur only in warm and humid climates. Besides this, our observations indicate that the disease is probably not all that infrequent among cats who are allowed to move about outdoors, considering that these two cases were picked up at a private, first opinion, small animal practice in Helsinki, both within a 5 year period. Recognising such cases is a matter of keeping the almost pathognomonic features in mind. Secondly, diagnostic verification requires a reference laboratory that is able to confirm the diagnosis correctly. The use of 16S rRNA gene sequencing helps in typing closely related mycobacteria. It was advantageous to do the imprints for cytology before dropping the piece of tissue into formalin. That made it possible quickly to identify mycobacteria-like acid-fast organisms with Ziehl-Neelsen staining. Another benefit of cytology over histopathology is due to the fact, that routine processes for histopathology include dehydration procedures with alcohol, which may cause the acid-fast organisms to stain poorly. Thus no mycobacteria could be detected in histopathology with Ziehl-Neelsen in either of our cases and only single bacteria were seen with Fite-Faraco stain.
Treating the panniculitis by using medical treatment only, turned out to be very tedious and frustrating. In the first case, the erroneous laboratory diagnosis and the lack of susceptibility testing from the start were obvious shortcomings. The combination of rifabutin at high doses and TMS seemed effective, but unfortunately the cat did not tolerate this medication very well. We assume that the concurrent folic acid helped the cat to tolerate the TMS for long periods. Rifabutin is not the drug of choice for infections by rapidly growing mycobacteria, and would probably never have been used in case one if a M avium-intracellulare type bacterium had not been erroneously diagnosed in the beginning. It is our opinion that this bacterium perhaps appeared at some point into the first sample as a contaminant. (On diagnosing the mycobacterium, seen with Ziehl-Neelsen on cytology, the laboratory had to work a lot with the sample to get rid of the disturbing fast growing component, which was in a majority and which at that point was considered to be a Nocardia sp.)
The waxing and waning course of the disease makes it difficult to judge when a treatment is effective. After a quiescent period of weeks or months, fresh fistulas often reappeared, not in the same region as before, but in nearby areas. It is interesting to note that during the follow up period of 7 years in the first case the infection never returned to the areas that once had healed well. It seemed to us as if a local immunity had developed over the years. Furthermore, the infection never spread to the neck or head, or the distal parts of the extremities. This could be due to lack of large amounts of adipose tissue in these areas. The degree of hyperproteinaemia did not seem to correlate with the advancement of the disease. High levels were also seen during calm periods and normal values also at times when the cat had numerous draining fistulas. There is a recent report of one cat with simultaneous infection by both a Nocardia species and an atypical mycobacterium and significantly elevated serum calcium concentration (Mealey et al 1999). During the first 3 years of infection our first case also had, at every blood test, serum calcium levels between 3.0 and 3.8 (reference 2.20–3.00 mmol/l). At a check up 7 years postdiagnosis, after having a calm period of about 2 years without medication, serum calcium concentration was normal (2.74 mmol/l).
It is surprising that the chronic inflammation and the heavy medication did not greatly affect the health of the cat. However, we have recently noted that the cat has severe retinal degeneration. Although over 2 years had elapsed from the last enrofloxacin medication to the first signs of visual impairment observed by the owner, we cannot exclude the possibility that this is a consequence of previous long-term enrofloxacin therapy at high dosages. Gelatt et al (2001) recently reported an association between rapid retinal degeneration and enrofloxacin administration in cats. This should receive attention in the future when prescribing long-term enrofloxacin to these cases, as immediate discontinuation of the medication in early stages of retinal intoxication might prevent total blindness. The cat in case two, which received injectable enrofloxacin, developed necrotic skin lesions of up to 1 cm2 at some of the injection sites (five sites) during the 6-week period of medication, probably from injections not very well placed within the subcutis.
Questions emerging from diagnosing and treating our two cases are: Why does panniculitis by M smegmatis seem to be more common in cats than in humans and dogs? Is it a question of how and where it is inoculated and is the immune defence of adipose tissue of cats different from that of humans and dogs?
It is our suspicion that infections by M smegmatis probably are more common outside Australia than previously considered. We have learnt that a vague laboratory diagnosis without susceptibility results is not of very much help. Confusing Nocardia, being partly acid-fast, with M smegmatis, having resembling microbiological features to Nocardia, hampers diagnostic work. To avoid this a laboratory that is specialised in difficult mycobacteria should be used.
Footnotes
Acknowledgements
We thank DVM Barbara Stein for an excellent presentation of mycobacterial panniculitis at the annual Veterinary Conference in Helsinki in November 1993. The authors want to thank Pirkko Koukila-Kähkölä for helpful suggestions in characterising the bacterium. Sini Suomalainen is acknowledged for DNA sequencing. DVM, PhD, Spec.SmallAnim.Dis. Jan Räihä has helped with proofreading this manuscript.
Addendum
After this manuscript was written the cat of case one was administered oral doxycycline monohydrate 50 mg/kg bid (DoximedR, Ratiopharm) for 7 months. The cat tolerated this medication well and the bumpy area on the back reduced in size considerably. After this only a small area right at the base of the tail feels slightly abnormal. Doxycycline should have been tried earlier ie, immediately the susceptibility results were received.